Set Of Landmarks Research Articles

Incorporating Compositions in Qualitative Approaches

Qualitative approaches to various tasks, ranging from localization and mapping to active planning, are gaining considerable momentum in recent years. These approaches represent the environment through spatial relationships between small sets of landmarks in independent local coordinate systems. An essential component in these approaches is the composition operator, enabling spatial information propagation between different sets to infer new ones. Integrating compositions within qualitative algorithms brings several difficulties. For instance, if the information required to perform a specific composition operation is unavailable, it must be inferred first, possibly via a preparatory composition operation. This recursive issue becomes more challenging as the amount of information grows. This letter addresses two main questions arising from the above, which remained open: 1. Given an initial set of qualitative spatial relationships, what new ones can be composed? 2. What is the optimal sequence of compositions operations to create a target set among all possible sequences? We provide a theoretical derivation to address the first question and a novel search algorithm to address the second.

A new approach to symmetric registration of longitudinal structural MRI of the human brain

BackgroundThis paper presents the Automatic Temporal Registration Algorithm (ATRA) for symmetric rigid-body registration of longitudinal T1-weighted three-dimensional MRI scans of the human brain. This is a fundamental processing step in computational neuroimaging. New methodThe notion of leave-one-out consistent (LOOC) landmarks with respect to a supervised landmark detection algorithm is introduced. An automatic algorithm is presented for identification of LOOC landmarks on MRI scans. Multiple sets of LOOC landmarks are identified on each volume and a Generalized Orthogonal Procrustes Analysis of the landmarks is used to find a rigid-body transformation of each volume into a common space where the volumes are aligned precisely. ResultsQualitative and quantitative evaluations of ATRA registration accuracy were performed using 2012 volumes from 503 subjects (4 longitudinal volumes/subject), and on a further 120 volumes acquired from 3 normal subjects (40 longitudinal volumes/subject). Since the ground truth registrations are unknown, we devised a novel method for showing that ATRA’s registration accuracy is at least better than 0.5 mm translation or 0.5° rotation. Comparison with existing method(s)In comparison with existing methods, ATRA does not require any image preprocessing (e.g., skull-stripping or intensity normalization) and can handle conditions where rigid-body motion assumptions are not true (e.g., movement in eyes, jaw, neck) and brain tissue loss over time in neurodegenerative diseases. In a systematic comparison with the FSL FLIRT algorithm, ATRA provided faster and more accurate registrations. ConclusionsThe algorithm is symmetric, in the sense that any permutation of the input volumes does not change the transformation matrices, and unbiased, in that all volumes undergo exactly one interpolation operation, which precisely aligns them in a common space. There is no interpolation bias and no reference volume. All volumes are treated exactly the same. The algorithm is fast and highly accurate.

Arthroscopic-assisted radiocarpal ligaments tensioning for dynamic radiocarpal instability

BackgroundDynamic radiocarpal instability is one of the causes of post-trauma radial-sided wrist pain. It is not easy to diagnose and may possibly be overlooked. The key ligaments responsible for dynamic radiocarpal instability are the radioscaphocapitate (RSC) and long radiolunate (LRL) ligaments. Tensioning of these 2 ligaments could be a method of treatment for dynamic carpal instability. We proposed a method for arthroscopic thermal shrinkage of these 2 ligaments, and for setting a landmark arthroscopically to facilitate identification of these 2 ligaments during the combined open suture tensioning procedure.MethodsBetween January 2016 and May 2020, 12 patients treated with this method were enrolled. The mean age was 33.3 years (range, 18–57 years), and the mean duration from injury to operation was 7.8 months (range, 3–25 months). The diagnosis was mainly depended on the physical examinations and confirmed under arthroscopy. The mean follow-up was 17.7 months (range, 12–26 months).ResultsAll the patients had marked improvement of pain, grip strength, the Disabilities of the Arm, Shoulder and Hand questionnaire (DASH), and the radiocarpal stability. The wrist range of motion showed significant decrease around 5o in both flexion and extension and around 4o in the ulnar deviation at the final follow-ups. All patients were able to return to their previous full level of work and activities.ConclusionsWe conclude that arthroscopic thermal shrinkage combined with open suture tensioning can be effective in treating dynamic carpal instability, while the arthroscopic-assisted landmark setting can help identify the accurate location of the RSC and LRL ligaments without dissecting too much soft tissue.

Morphological correspondence between brain and endocranial surfaces in mice exposed to undernutrition during development.

The morphological changes of the brain and the skull are highly integrated as a result of shared developmental pathways and different types of interactions between them. Shared developmental trajectories between these two structures might be influenced by genetic and environmental factors. Although the effect of environmental factors on neural and craniofacial traits has been extensively studied, less is known about the specific impact of stressful conditions on the coordinated variation between these structures. Here, we test the effect of early nutrient restriction on morphological correspondence between the brain and the endocast. For this purpose, mice exposed to protein or calorie-protein restriction during gestation and lactation were compared with a control group in which dams were fed standard food ad libitum. High-resolution images were obtained after weaning to describe brain and endocranial morphology. By magnetic resonance imaging (MRI), brain volumes were obtained and endocasts were segmented from skull reconstructions derived from micro-computed tomography (microCT). Brain and endocranial volumes were compared to assess the correspondence in size. Shape changes were analyzed using a set of landmarks and semilandmarks on 3D surfaces. Results indicated that brain volume is relatively less affected by undernutrition during development than endocast volume. Shape covariation between the brain and the endocast was found to be quite singular for protein-restricted animals. Procrustes distances were larger between the brain and the endocast of the same specimens than between brains or endocasts of different animals, which means that the greatest similarity is by type of structure and suggests that the use of the endocast as a direct proxy of the brain at this intraspecific scale could have some limitations. In the same line, patterns of brain shape asymmetry were not directly estimated from endocranial surfaces. In sum, our findings indicate that morphological variation and association between the brain and the endocast is modulated by environmental factors and support the idea that head morphogenesis results from complex processes that are sensitive to the pervasive influence of nutrient intake.

How accurate is forensic facial identification of surgically altered faces?

ABSTRACT Surgical alterations of the face have been shown to be a defining factor in alterations of physical appearance. It is estimated that 5% of the population undergo corrective procedures of the jaws for aesthetic or health purposes. To date, however, practitioners have been lacking guidelines specifying the way facial identification should be conducted. The present study explores the effect of oral and maxillofacial surgeries on landmark-based algorithms when conducted on 3D imagery. The sample composed of 50 pre-op and post-op 3D face images of individuals who had undergone corrective surgeries was tested against a set of 500 faces from the general European population. An approach based on discrete points was employed while the number of landmarks was gradually reduced to minimize intra-person and maximize inter-person variance. The results showed that the presence of surgical modifications resulted in misclassifications based on the mid-face and lower face features, particularly upper lip protrusion and shape, lower lip protrusion, and chin protrusion. The optimal set of landmarks which avoided the surgically affected regions reached an accuracy of 92% at the rank of 1% and 98% at the rank of 5. Across all tested landmark configurations, males were consistently less prone to mismatches than females.

Facial Expressions Recognition Based on Delaunay Triangulation of Landmark and Machine Learning

Facial expressions can tell a lot about an individual’s emotional state. Recent technological advances opening avenues for automatic Facial Expression Recognition (FER) based on machine learning techniques. Many works have been done on FER for the classification of facial expressions. However, the applicability to more naturalistic facial expressions remains unclear. This paper intends to develop a machine learning approach based on the Delaunay triangulation to extract the relevant facial features allowing classifying facial expressions. Initially, from the given facial image, a set of discriminative landmarks are extracted. Along with this, a minimal landmark connected graph is also extracted. Thereby, from the connected graph, the expression is represented by a one-dimensional feature vector. Finally, the obtained vector is subject for classification by six well-known classifiers (KNN, NB, DT, QDA, RF and SVM). The experiments are conducted on four standard databases (CK+, KDEF, JAFFE and MUG) to evaluate the performance of the proposed approach and find out which classifier is better suited to our system. The QDA approach based on the Delaunay triangulation has a high accuracy of 96.94% since it only supports non-zero pixels, which increases the recognition rate.

Technical note: The effect of image annotation with minimal manual interaction for semiautomatic prostate segmentation in CT images using fully convolutional neural networks.

The goal is to study the performance improvement of a deep learning algorithm in three-dimensional (3D) image segmentation through incorporating minimal user interaction into a fully convolutional neural network (CNN). A U-Net CNN was trained and tested for 3D prostate segmentation in computed tomography (CT) images. To improve the segmentation accuracy, the CNN's input images were annotated with a set of border landmarks to supervise the network for segmenting the prostate. The network was trained and tested again with annotated images after 5, 10, 15, 20, or 30 landmark points were used. Compared to fully automatic segmentation, the Dice similarity coefficient increased up to 9% when 5-30 sparse landmark points were involved, with the segmentation accuracy improving as more border landmarks were used. When a limited number of sparse border landmarks are used on the input image, the CNN performance approaches the interexpert observer difference observed in manual segmentation.

Link dimension and exact construction of graphs from distance vectors

Minimum resolution set and associated metric dimension provide the basis for unique and systematic labeling of nodes of a graph (G) using distances to a set of landmarks. Such a distance vector set, however, may not uniquely identify the graph or allow for its exact construction. The concept of construction set is presented, which overcomes these limitations. Link dimension (ξ(G)) is the minimum cardinality of a construction set. We study ambiguity in the given distance vector set and conditions for resolving them. Bounds over link dimension, proofs for link dimension for various graph families and relationship between graph dimensions and graph spectra are presented. The inverse problem of identifying whether a given set of non-negative distance vectors can be associated with a graph is also addressed, introducing the concept of graphability of distance vectors. A graph of N nodes can now be compressed to a representation with an N×ξ(G) matrix which allows for its exact reconstruction.

Structure-Driven Unsupervised Domain Adaptation for Cross-Modality Cardiac Segmentation

Performance degradation due to domain shift remains a major challenge in medical image analysis. Unsupervised domain adaptation that transfers knowledge learned from the source domain with ground truth labels to the target domain without any annotation is the mainstream solution to resolve this issue. In this paper, we present a novel unsupervised domain adaptation framework for cross-modality cardiac segmentation, by explicitly capturing a common cardiac structure embedded across different modalities to guide cardiac segmentation. In particular, we first extract a set of 3D landmarks, in a self-supervised manner, to represent the cardiac structure of different modalities. The high-level structure information is then combined with another complementary feature, the Canny edges, to produce accurate cardiac segmentation results both in the source and target domains. We extensively evaluate our method on the MICCAI 2017 MM-WHS dataset for cardiac segmentation. The evaluation, comparison and comprehensive ablation studies demonstrate that our approach achieves satisfactory segmentation results and outperforms state-of-the-art unsupervised domain adaptation methods by a significant margin.

Three-Dimensional Cephalometric Landmarking and Frankfort Horizontal Plane Construction: Reproducibility of Conventional and Novel Landmarks.

In some dentofacial deformity patients, especially patients undergoing surgical orthodontic treatments, Computed Tomography (CT) scans are useful to assess complex asymmetry or to plan orthognathic surgery. This assessment would be made easier for orthodontists and surgeons with a three-dimensional (3D) cephalometric analysis, which would require the localization of landmarks and the construction of reference planes. The objectives of this study were to assess manual landmarking repeatability and reproducibility (R&R) of a set of 3D landmarks and to evaluate R&R of vertical cephalometric measurements using two Frankfort Horizontal (FH) planes as references for horizontal 3D imaging reorientation. Thirty-three landmarks, divided into “conventional”, “foraminal” and “dental”, were manually located twice by three experienced operators on 20 randomly-selected CT scans of orthognathic surgery patients. R&R confidence intervals (CI) of each landmark in the -x, -y and -z directions were computed according to the ISO 5725 standard. These landmarks were then used to construct 2 FH planes: a conventional FH plane (orbitale left, porion right and left) and a newly proposed FH plane (midinternal acoustic foramen, orbitale right and left). R&R of vertical cephalometric measurements were computed using these 2 FH planes as horizontal references for CT reorientation. Landmarks showing a 95% CI of repeatability and/or reproducibility > 2 mm were found exclusively in the “conventional” landmarks group. Vertical measurements showed excellent R&R (95% CI < 1 mm) with either FH plane as horizontal reference. However, the 2 FH planes were not found to be parallel (absolute angular difference of 2.41°, SD 1.27°). Overall, “dental” and “foraminal” landmarks were more reliable than the “conventional” landmarks. Despite the poor reliability of the landmarks orbitale and porion, the construction of the conventional FH plane provided a reliable horizontal reference for 3D craniofacial CT scan reorientation.

Comparing and combining sliding semilandmarks and weighted spherical harmonics for shape analysis.

Quantifying morphological variation is critical for conducting anatomical research. Three-dimensional geometric morphometric (3D GM) landmark analyses quantify shape using homologous Cartesian coordinates (landmarks). Setting up a high-density landmark set and placing it on all specimens, however, can be a time-consuming task. Weighted spherical harmonics (SPHARM) provides an alternative method for analyzing the shape of such objects. Here we compare sliding semilandmark and SPHARM analyses of the calcaneus of Gorilla gorilla gorilla (n=20), Pan troglodytes troglodytes (n=20), and Homo sapiens (n=20) to determine whether the SPHARM and sliding semilandmark analyses capture comparable levels of shape variation. We also compare both the sliding semilandmark and SPHARM analyses to a novel combination of the two methods, here termed SPHARM-sliding. In SPHARM-sliding, the vertices of the surface models produced from the SPHARM analysis (that are the same in number and relative location) are used as the starting landmark positions for a sliding semilandmark analysis. Calcaneal shape variation quantified by all three analyses was summarized using separate principal components analyses. Results were compared using the root mean square (RMS) and maximum distance between surface models of species averages scaled (up) to centroid size created from each analysis. The average RMS was 0.23mm between sliding semilandmark and SPHARM average surface models, 0.19mm between SPHARM and SPHARM sliding average surface models, and 0.22mm between sliding semilandmark and SPHARM sliding average surface models. Although results indicate that all three analyses are comparable methods for 3D shape analysis, there are advantages and disadvantages to each. While the SPHARM analysis is less time-intensive, it is unable to capture the same level of detail around the sharp edges of articular facets on average surface models as the sliding semilandmark analysis. The SPHARM analysis also does not allow for individual articular facets to be analyzed in isolation. SPHARM-sliding, however, captures the same level of detail as the sliding semilandmark analysis, and (as in the sliding semilandmark analysis) allows for the evaluation of individual portions of bone. SPHARM is a comparable method to a 3D GM analysis for small, irregularly shaped bones, such as the calcaneus, and SPHARM-sliding allows for an expedited set up process for a sliding semilandmark analysis.

Accuracy of Virtual Surgical Planning with Mandible-first Surgery and Factors that may Affect It

Planning bimaxillary orthognathic surgery has become much easier and the workflow more streamlined with the advent of virtual surgical planning (VSP). Among its advantages, VSP allows for easy transition between performing maxillary or mandibular surgery first, as it eliminates the cumbersome process of reverse model surgery. In this experience of performing mandible surgery first, it was observed that the osteotomy gaps at the piriform and zygoma appeared to be different than what was predicted on the VSP. The present study was initiated to compare the planned movements to the actual achieved surgical position of the maxilla after correction of dentoskeletal deformity. The goal of the study was to evaluate the accuracy of a maxillary surgery-first sequence and a mandibular surgery-first sequence. Additionally, the researchers sought to determine if any surgical interventions could mitigate the under-advancement noted in observations.The authors implemented a retrospective audit of 61 patients who underwent bimaxillary orthognathic surgery from January 2017 to September 2020. Exclusion criteria included previous cleft lip/palate surgery, previous orthognathic surgery, lack of a post-surgical CBCT, non-sagittal split osteotomies of the mandible, or if there was greater than 3mm vertical deviation in planned and achieved surgical position of the central incisor based on intraoperative incisor show. Patients were divided into 2 groups based on the surgical sequencing used; 28 patients had maxillary surgery performed first and 33 underwent mandibular-first surgery. Pre-surgical and post-surgical CBCT scans were analyzed by 3D Systems engineers, with overlays made using the predicted osteotomies and the postoperative CBCT scans. A pre-determined set of skeletal landmarks were measured on the virtual surgical plan and the actual postoperative images, and the difference between the planned and actual position was calculated for each reference point for each patient. The absolute value of each difference was then calculated, and a series of 2 sample T-tests was performed to compare the average absolute difference between planned and actual position between groups.The absolute values of the linear (mm) and angular (°) measurement differences between planned and postoperative results demonstrate statistically significant differences in accuracy between the 2 groups, with greater deviation in the mandible-first group. The maxillary central incisors were under advanced in the anterior-posterior (AP) direction in both groups. Additionally, a significant difference in reference points ISU1 (AP) and A Point (AP) was noted between the groups, indicating that greater inaccuracy lies within the mandibular-first group. However, the majority of data points show deviation from the surgical plan less than or equal to 2mm and 4 degrees, meeting previously described success criteria.1,2To further evaluate the larger maxillary under advancement in the mandible-first surgery cohort, the researchers sub-divided the mandible-first surgery group to evaluate factors that may explain the difference, 1 factor being the type of fixation used for sagittal split segments, rigid or non-rigid. When comparing the average linear measurements, the non-rigid fixation group showed less accuracy when compared to the rigid fixation group, and this was statistically significant (P-value: .007).In conclusion, the findings of the present study demonstrate that virtual surgical planning can be less accurate in predicting the maxillary incisor position when mandibular surgery is performed first, but this inaccuracy is within the acceptable range and can be mitigated by more rigid fixation of the mandible.

S109 Pancreatic Insufficiency and Time to Disease Progression Following Immune Checkpoint Inhibitor Pancreatic Injury in Advanced Cancers

Introduction: Immune checkpoint inhibitor-induced pancreatic injury (ICI-PI) ranges in presentation from asymptomatic enzyme elevation to symptomatic acute pancreatitis. Long-term outcomes of these patients remain poorly defined. This study characterizes progression-free survival and development of exocrine and endocrine pancreatic insufficiency after ICI-PI. Methods: We retrospectively queried all cancer patients at our institution who received one or more doses of cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) inhibitor, programmed cell death-1 (PD-1)/PD-1 ligand (PD-L1) inhibitor or both between 2011 and 2019 and later had at least one serum lipase elevation. Two physicians including a pancreatologist assessed cases to confirm ICI-induced pancreatic injury after ruling out all other etiologies. Exocrine pancreatic insufficiency was defined by fecal elastase (< 200ug/g) and new need for pancreatic enzyme supplementation. Endocrine insufficiency was defined by new-onset diabetes mellitus. Results: Of 6450 patients treated with ICI, 364 (6%) had elevated serum lipase while on ICI therapy. Of these, 105 (29%) had ICI-PI of which 12 (11%) developed pancreatic insufficiency: three (3%) with exocrine insufficiency and 9 (9%) with endocrine insufficiency over a median follow-up of 673 days. All patients with exocrine insufficiency and four with endocrine insufficiency occurred among the 27 (26%) ICI-PI patients that met Atlanta criteria for acute pancreatitis. No patient with pancreatic insufficiency had pre-ICI hyperlipasemia, pancreatic metastasis, abdominal radiation, or evidence of chronic pancreatitis on subsequent scans. ICI-PI patients manifesting as acute pancreatitis were significantly more likely than ICI-PI patients without acute pancreatitis to have pancreatic exocrine insufficiency (11% vs 0%, P=0.02) and persistent/recurrent hyperlipasemia (56% vs 17%, p< 0.01) but not endocrine insufficiency (15% vs 6%, P=0.23; Table 1). As compared with non-ICI injury, patients with any ICI-PI had 49% lower hazard of disease progression (HR 0.51, 95% CI:0.35-0.76, p< 0.01) and longer median time to progression (41.7 vs 14.3 months) (Figure 1). Patients with ICI-induced acute pancreatitis had trend toward reduction in hazard of disease progression compared to patients without pancreatitis (HR 0.63, 95% CI:0.38-1.07, P=0.08). Conclusion: Patients with acute ICI pancreatitis should be monitored for exocrine pancreatic insufficiency. The presence of any ICI-PI may portend lower hazard of disease progression.Table 1.: Long-Term Outcomes Following Immune Checkpoint Inhibitor (ICI)-induced Pancreatic Injury.Figure 1.: Kaplan-Meier time to disease progression stratified by cause of pancreatic injury. Conditional landmark set at 4 months from ICI initiation to reduce potential for guarantee-time bias.

MAXSAT Heuristics for Cost Optimal Planning

The cost of an optimal delete relaxed plan, known as h+, is a powerful admissible heuristic but is in general intractable to compute. In this paper we examine the problem of computing h+ by encoding it as a MAXSAT problem. We develop a new encoding that utilizes constraint generation to support the computation of a sequence of increasing lower bounds on h+. We show a close connection between the computations performed by a recent approach for solving MAXSAT and a hitting set approach recently proposed for computing h+. Using this connection we observe that our MAXSAT computation can be initialized with a set of landmarks computed by LM-cut. By judicious use of MAXSAT solving along with a technique of lazy heuristic evaluation we obtain speedups for finding optimal plans over LM-cut on a number of domains. Our approach enables the exploitation of continued progress in MAXSAT solving, and also makes it possible to consider computing or approximating heuristics that are even more informed that h+ by, for example, adding some information about deletes back into the encoding.

Automated description of the mandible shape by deep learning

PurposeThe shape of the mandible has been analyzed in a variety of fields, whether to diagnose conditions like osteoporosis or osteomyelitis, in forensics, to estimate biological information such as age, gender, and race or in orthognathic surgery. Although the methods employed produce encouraging results, most rely on the dry bone analyses or complex imaging techniques that, ultimately, hamper sample collection and, as a consequence, the development of large-scale studies. Thus, we proposed an objective, repeatable, and fully automatic approach to provide a quantitative description of the mandible in orthopantomographies (OPGs).MethodsWe proposed the use of a deep convolutional neural network (CNN) to localize a set of landmarks of the mandible contour automatically from OPGs. Furthermore, we detailed four different descriptors for the mandible shape to be used for a variety of purposes. This includes a set of linear distances and angles calculated from eight anatomical landmarks of the mandible, the centroid size, the shape variations from the mean shape, and a group of shape parameters extracted with a point distribution model.ResultsThe fully automatic digitization of the mandible contour was very accurate, with a mean point to the curve error of 0.21 mm and a standard deviation comparable to that of a trained expert. The combination of the CNN and the four shape descriptors was validated in the well-known problems of forensic sex and age estimation, obtaining 87.8% of accuracy and a mean absolute error of 1.57 years, respectively.ConclusionThe methodology proposed, including the shape model, can be valuable in any field that requires a quantitative description of the mandible shape and a visual representation of its changes such as clinical practice, surgery management, dental research, or legal medicine.

Improving a Rapid Alignment Method of Tomography Projections by a Parallel Approach

The high resolution of synchrotron cryo-nano tomography can be easily undermined by setup instabilities and sample stage deficiencies such as runout or backlash. At the cost of limiting the sample visibility, especially in the case of bio-specimens, high contrast nano-beads are often added to the solution to provide a set of landmarks for a manual alignment. However, the spatial distribution of these reference points within the sample is difficult to control, resulting in many datasets without a sufficient amount of such critical features for tracking. Fast automatic methods based on tomography consistency are thus desirable, especially for biological samples, where regular, high contrast features can be scarce. Current off-the-shelf implementations of such classes of algorithms are slow if used on a real-world high-resolution dataset. In this paper, we present a fast implementation of a consistency-based alignment algorithm especially tailored to a multi-GPU system. Our implementation is released as open-source.

Embryonic development in the acoel Hofstenia miamia.

Acoels are marine worms that belong to the phylum Xenacoelomorpha, a deep-diverging bilaterian lineage. This makes acoels an attractive system for studying the evolution of major bilaterian traits. Thus far, acoel development has not been described in detail at the morphological and transcriptomic levels in a species in which functional genetic studies are possible. We present a set of developmental landmarks for embryogenesis in the highly regenerative acoel Hofstenia miamia. We generated a developmental staging atlas from zygote to hatched worm based on gross morphology, with accompanying bulk transcriptome data. Hofstenia embryos undergo a stereotyped cleavage program known as duet cleavage, which results in two large vegetal pole 'macromeres' and numerous small animal pole 'micromeres'. These macromeres become internalized as micromere progeny proliferate and move vegetally. We also noted a second, previously undescribed, cell-internalization event at the animal pole, following which we detected major body axes and tissues corresponding to all three germ layers. Our work on Hofstenia embryos provides a resource for mechanistic investigations of acoel development, which will yield insights into the evolution of bilaterian development and regeneration.

Independent mobility for persons with VIB using GIS

Purpose The study aims to promote independent mobility for persons with visual impairment or legal blindness (VIB) by developing a Web-based wayfinding application using geographic information systems (GIS). While the literature mainly focuses on technical devices presenting results for wayfinding, a lack of integration of user needs is identified. The inclusive, participative developed application offers step-by-step directions for pre-trip planning through an accessible user interface. Design/methodology/approach The paper presents a semi-automated approach to extract a pedestrian routing network data set based on open government data and field survey. User profiles calculate different routes using a weighting scheme for landmarks, orientation hints, infrastructure of crossings and sidewalks. The adoption of ArcGIS Web AppBuilder widgets allows access of the user interface additionally through keyboard-only navigation and therefore screen-reader capability. Findings GIS offers a powerful tool to design network analysis for persons with VIB. The routing algorithm accesses different user profiles, returning individualized turn-by-turn directions. The complex set of attributes, including shorelines, landmarks and barriers, can be integrated by semi-automated processes. Practical implications The paper illustrates the benefit of GIS applications for wayfinding of persons with VIB to raise self-determination and independence. Originality/value A ubiquitous pedestrian sidewalk network for a medium-size city comprises a novelty, as research has mainly focused on small areas. The integration of shorelines next to a various number of hints, landmarks and potential barriers through semi-automated processes allows reproducibility and transferability of the model to other cities.

Estimating mean potential outcome under adaptive treatment length strategies in continuous time.

An adaptive treatment length strategy is a sequential stage-wise treatment strategy where a subject's treatment begins at baseline and one chooses to stop or continue treatment at each stage provided the subject has been continuously treated. The effects of treatment are assumed to be cumulative and, therefore, the effect of treatment length on clinical endpoint, measured at the end of the study, is of primary scientific interest. At the same time, adverse treatment-terminating events may occur during the course of treatment that require treatment be stopped immediately. Because the presence of a treatment-terminating event may be strongly associated with the study outcome, the treatment-terminating event is informative. In observational studies, decisions to stop or continue treatment depend on covariate history that confounds the relationship between treatment length on outcome. We propose a new risk-set weighted estimator of the mean potential outcome under the condition that time-dependent covariates update at a set of common landmarks. We show that our proposed estimator is asymptotically linear given mild assumptions and correctly specified working models. Specifically, we study the theoretical properties of our estimator when the nuisance parameters are modeled using either parametric or semiparametric methods. The finite sample performance and theoretical results of the proposed estimator are evaluated through simulation studies and demonstrated by application to the Enhanced Suppression of the Platelet Receptor IIb/IIIa with Integrilin Therapy (ESPRIT) infusion trial data.

Deep Unsupervised 3D Human Body Reconstruction from a Sparse set of Landmarks

In this paper we propose the first deep unsupervised approach in human body reconstruction to estimate body surface from a sparse set of landmarks, so called DeepMurf. We apply a denoising autoencoder to estimate missing landmarks. Then we apply an attention model to estimate body joints from landmarks. Finally, a cascading network is applied to regress parameters of a statistical generative model that reconstructs body. Our set of proposed loss functions allows us to train the network in an unsupervised way. Results on four public datasets show that our approach accurately reconstructs the human body from real world mocap data.